# distribution by stariya

VIEWS: 11 PAGES: 8

• pg 1
```									                          Lab 6: Trip Distribution with TRANSCAD
CE 451/551

OBJECTIVES:
 Examine the gravity model and the input information required,
 Apply the gravity model to a small scale transportation network,
 Develop an understanding of the gravity model and how it fits into the TDM process,
 Understand the outputs of the gravity model.

DATA: Click on the link to Files in the syllabus and save the file to your working folder. Uncompress
the files. These are the TransCAD files for an older version of the Ames travel demand model.

Report: No comprehensive write-up required. You may cut and paste from this document to prepare your
type - not in boldface.) Make sure plots and graphs are neatly done, are labeled properly and are readable,
especially if you use print-screen copy/paste. Use appropriate significant figures in results. Don’t forget
TransCAD has online help (F1) if you need to find out how to do something you can’t remember or have
not seen yet.

Production and Attraction Estimates

Trip productions and attractions for the 1995 Ames area model are provided in the worksheet
(AmesP&A.xls). The Ps and As were based on demographic (STF1) and socio-economic (STF3) data
from Census data and Employment Data from the Department of Workforce Development (then known as
the Department of Employment Services or DES), aggregated to the zonal (TAZ) level. The original Trip
generation computations are also included. Open and Review that worksheet to assess how the
Productions and Attractions were developed. A portion of the equations are shown below, along with
related questions you should review (and answer) before we proceed to use the values in trip distribution:

Trip Generation Formulas used in Ames

The excel database consists of data fields (columns) from the DES (numbers of employees in
various categories, e.g., retail, service and other). It also includes fields from the Census at two
geographies, block (STF1, demographic data, such as the number of households, and the number
of households of various sizes, e.g., 1 person, 2 person, …) and block group (STF3,
socioeconomic data, such as the number of households in each income group 0-20K, 20-40K, and
>40K).

1. Write out the equation used for HBW Productions. Do not use excel cell notation in your
equation, rather, use variable names. Important: use the equation in row 7, cell AG7 (row 6 has
some zeros that may affect your understanding of the equation.)

For example, your equation should look similar to this:

0.16*{3.7*NL* + 6.8*NL2 …) + 0.21*{ 4.6*NM1 + …} + 0.20* {…}

1
Where

NL1 = number of low income, one person households
NL2 = number of low income, two person households
NM1 = number of medium income, one person households…and so forth.

Since the variables on the right side of the equation are also products of equations themselves,
write out the equation for the first of them (NL1). You needn’t write out the equation for NL2,
etc., as they would be similar. Make sure your equation for NL1 has only data fields on the right
hand side (blue fields), not variables that have been calculated from data (white or yellow fields).

Note: TAZ #1 is ISU, originally modeled as a single zone. The “modified” section (starting on row 81)
of the spreadsheet shows how zone 1 was split up into several sub-zones (TAZs 1, 69 and 70) – see
column AR. Zones 3, 5 and 59 were also modified somewhat due to missing data in the Census. For the
“future” mode scenario (starting on row 170), most zones were assumed to have the same SE
characteristics. Purple cells indicate cells are for zones where growth is expected to occur (see also
column AR). Zones 72-78 are external zones (zones representing trips that leave, enter or pass through
the Ames area.)

2. Does this appear to be a cross-classification model or a regression model? Why?

Note that NHB Productions are set to = NHB Attractions, while for other purposes, we generate trip
productions and attractions independently.

Trip attraction equations were also developed for HBW, HBO and NHB trips.

3. What independent variables were used in the attraction models?

P and A format for Application (read this section and examine the GM.dbf file in TransCAD or Excel –
note, if you use Excel, you will need to right click on the dbf file, and choose open with – choose program
and find the excel program which is listed under C:\Program Files\Microsoft Office\Office12).

Before the production and attraction data can be used in the TransCAD software, the analyst must put the
productions and attractions for each TAZ on the same line or record in a database. An easy way to
accomplish this is to create a separate worksheet or Excel file. The data can then be cut and pasted so
each TAZ has P’s and A’s for each trip purpose together. The Excel file or tab can then be exported as a
comma delimited format (CSV) file and then opened easily in the TransCad software program. Once the
.csv file is open in TransCad it can be saved as a dbf file (note that the new version of Excel cannot save
to dbf format directly).

Important: Remember – green background means the data are read only. Closing the dbase file and
reopening it in TransCAD (uncheck the read only bit) enables editing. It can easily be joined to the
existing line layer (nodes) based on the unique TAZ identifier. (Centroid numbers of the nodes will
match the TAZ numbers.) An example of the dataview is shown below. To save some steps, a dbf file
with the productions and attractions contained has been completed for you and saved as GM.dbf.
Examine that file to see the final product.

2
Friction Factors

Friction factor table. Friction factors are a measure of travel impedance in the model. They are used in
the gravity model equation to help determine trip lengths for each type of trip purpose. The friction
factors for the Ames model were borrowed from another model region, a common practice. The original
friction factor table was available as a text file and has been converted to be in the form needed by
TransCAD and saved as a Friction.dbf.

4. Open Friction .dbf with Excel and develop line graphs showing the HBW and HBO factors
on the same chart. Print the chart in your lab report. Since trips in Ames are typically 20
minutes or less (except by transit), plot only the range of friction values for the time increments
from 1 to 20 minutes.

5. From the chart, draw a conclusion about the expected average trip lengths, relative to each
other for the two trip types, based on the shape of the friction factor functions.

Close the Excel file of friction factors. Start TransCAD and Open the STREETS map (if prompted,
choose “OK to update to current version” as the original was not saved in TransCAD v. 5.0).

Centroids

6. On the Nodes/Intersections Layer Dataview, what would we call the set of rows (nodes) that
are currently selected (red dots) e.g., what do you think these nodes represent? Close the
dataview.

7. Highlight the CENTROID layer in the “Working Layer” window and open the data view. How
could you change the labels on the map from the current values to the values that represent
the TAZ numbers? List the command you should use and the settings for that command
such that TAZ numbers will be the labels on the map. Notice that the ID label in this case is
not the label you want to use.

3
Balancing Trips

Step through the process of balancing the productions and attractions in the GM.dbf file. Follow the
procedure in the short Chapter 4 of Manual B to incorporate all three types simultaneously. Pay particular
attention to the “To do this… Do this…” sections. Make sure you are balancing the correct attraction and
production vectors (careful … TransCAD tries to be helpful but default values may be wrong!) Home-
based trips should be balanced using the production vector as the control. Non-home-based trips should
use the attraction vector as the control. The output will automatically save a BALANCE.bin (binary
file). This file will be needed later for the gravity model. A combined file with the balanced and
unbalanced data will be produced during the process. Save this file as a .dbf (balance.dbf) so you can
explore the results and use it later.

Use the “Compute Statistics” tool to compare the balanced trips. An example is shown below (note:
PHBW are the original HBW productions and PHBW1 are the balanced productions, etc.):

8. List the average values before and after balancing for each of the trip purposes’ Ps and As.

Gravity Model Application

In this section, you will use Network/Paths>Multiple paths to do a Multipath analysis and create a shortest
path matrix that shows the minimum travel time between all zones. This minimum time path tree will be
used to run the gravity model application. You will want the Highway/Streets Layer open as the active
layer, and you will be minimizing travel time. The system stores the shortest path matrix in the default
file SPMAT.MTX.

   First, create a network. Have Highways/Streets as the Active Layer, then, from the Network/Paths

   Choose Create. “Length” should be selected as the length field. Add “time” as a link field

4
   Click OK and save the network as lab6.net – the .net will append automatically, so all you have to
type is lab6 in the filename field.) Remember, the network is not the same thing as the line layer.
Although it may use all the links in the line layer, it doesn’t have to.

Any time the network file is modified, the modeler would proceed to “update the network” (to make sure
changes are reflected in the network as well as the line file). To update the network you would use the
“Networks/Paths> Settings” dialog box (open that and have a look). Here you can also add or change turn
penalties/global u-turn prevention/ and prevent TransCad from using centroid connectors as through
routes. Before using this dialog box you would first want to create a selection set of the “centroids” in
your network. Then as you go through the "Networks/Paths>Settings” dialog box, click on the "Create
from selection set" radio button and select "Centroids". Then you’d select "turn penalties" and specify no
u-turns etc. You do not need to do those steps for this lab, but you will in future labs.

   Next, make the Nodes/Intersection layer active and make a selection set of all the centroids (hint:
note that centroids are all numbered ID<100).

   Make the Highway/Streets layer active.

   Shortest Path matrix: With the Nodes or Streets Layer Active, From Network/Paths Menu,
choose Networks/Paths - Multiple Paths and minimize “time.” The minimum travel time should
be calculated from “Selection” to “Selection”. The selection set in this case is the centroids in the
model. (The Multiple Paths function creates the skim trees and travel time matrix from all nodes
to all nodes, or from selected nodes like centroids, for the network). Check the radio button
“Matrix” to save impedance values as a matrix. Click OK and save the output matrix as lab6 in
your working folder (note that it might not default to that folder). See figures on next page and
note the magnitude of trips from cell to cell – some fairly small flows, and fractions of vehicle
trips, at that!

The result from this operation does not include
any intrazonal travel times. Realistically, the
diagonal elements should have some value.

5
TransCAD offers an easy strategy for filling in the diagonals.

   With the SPMAT.MTX open, you can use the Matrix pulldown menu “fill…”. Use a Fill Value of
1 (1 minute) and check the Radio Button for Diagonal. All diagonal values would then have a
non-zero value. If larger values were needed in some cells you could manually adjust those
values.

9. What is the TransCAD function/command that you could use to automatically fill the
diagonal with more meaningful values? Hint – the command is under Planning/Trip
Distribution. Run this tool with its default values.

10. What is the new intrazonal travel time for zone 5?

Now you are ready to apply the gravity model. In our case we have assumed the friction factors have
been calibrated for the study area. Following the tutorial in Manual B (Page 77), apply the gravity model
to each of the three trip purposes using the appropriate friction factors. Steps follow …

The following assumes you have obtained a balanced production and attraction .dbf file and the Ames
Street map is open.

Open the Friction factor table discussed above (a dbf file). The file should have Time and the FF for
three purposes, HBW, HBO and NHB.

Gravity Model application

At this time you should have the Ames map open with a network, a matrix with the shortest paths, and the
Friction Table open. Also open GM+balance.dbf..

   from the Planning Menu, choose Trip Distribution, and Gravity Application. Before proceeding,
make sure in the Production-Attraction Data part of the form, the “Table” is set to “GM+Balance”
and on the Friction Factor Settings part of the form, “FF Table” is set to “FRICTION” and be sure
the “impedance matrix” is the “shortest path matrix.” If several matrices are open the program
may pick a file that you may not want. Check to be sure the constraint type is “doubly
constrained.”

   In the “General” part of the form, you should add the three trip types from your balanced P and A
file (red pencil icon). Type in each purpose (default is purpose n – change to HBW, e.g.,).
Choose the appropriate balanced production column (e.g., PHBW1 for HBW) and the appropriate
Attraction Column should automatically be brought in. Note: you must click the "Add (red
pencil)" button to add another trip purpose.

   In the “Friction Factor” part of the form, for each trip purpose, identify the column with the FF for
the purpose you are evaluating.

6


   Click the settings tool. This is important as it lets you save your settings, which saves time if you
make a mistake or want to run the procedure again in the future. Save your settings (green plus
sign). Note: that if you come back to do another gravity model, you can use the settings button
and not have to enter all the settings by hand. Click OK

   Click OK to Run the Gravity Application. Give your settings file a unique name (e.g., lab6PA) so
you can identify it later (the default name is CGRAV.MTX). Click “show warnings” and note the
last section to answer the next question:

11. How many iterations did the gravity model take to converge? (note: this is a more meaningful
question after you have done your homework on the gravity model)

Combine output matrices - The output so far is a trip distribution table or matrix (Gravity Matrix) that has
3 layers (one for each of three trip purposes). Combine the three trip types using Matrix>QUICKSUM to
sum the values into a total trip interchange layer (called “QuickSum). Note the large number of intrazonal
trips in zone 1.

12. What type of land use do you think is in zone 1?

Determine the maximum, minimum and average trip lengths for each trip purpose:

With the shortest path matrix (spmat) and the Gravity Matrix (cgrav or appropriate name) open, go to
7
“Planning>Trip Distribution>Trip Length Distribution”

The base matrix file is the GRAVITY MATRIX and the MATRIX box should be used to choose trip
purpose(s) you want to analyze (default is QUICKSUM, but you should choose HBW or purpose 1). The
Impedance Matrix is the shortest path matrix, and you should use shortest path - travel time. Click
"Options". From there click the radio button "Bins start at" and input "0". Select "End at" and input 25.
Set number of Bins is equal to "5". Click OK twice, and save the "tld.mtx" file by clicking OK. Do this
for each trip purpose (read questions 15 and 16 before doing the other trip purposes).

13. For each purpose, note from the popup window the Minimum/Maximum/Average and Std.
Deviation of the trip length. Make the charts legible.

14. For each purpose, click OK and plot the model Trip Length Frequency Distribution.

15. Try the Shortest Path tool to check the reasonableness of routing in your network, by selecting 5
pairs of zones in which one zone is “across town” from the other. Plot each route and note its
travel time on a map. Make you maps legible.

16. What could be the network-based reasons for any illogical routing that might be observed (if
you do not observe any illogical routing, speculate what might cause the model to created
illogical routes)?

8

```
To top